In-vitro human lung cell injuries induced by urban PM2.5 during a severe air pollution episode: Variations associated with particle components

Environmental air pollutants pose significant threats to public health, especially the toxicity and diseases caused by the atmospheric fine particulate matters (PM2.5). Since the health risks vary with both the concentrations and compositions of PM2.5 which are determined by aerosol sources, how are their toxic effects relevant to the pollution level becomes an important issue, such as the haze episodes covering clean and polluted days. With the transition from non-pollution to pollution stage, daily PM2.5 samples were collected from both the urban and industrial areas of Nanjing city, eastern China, covering a typical haze event in autumn-winter. Their unpropitious effects on human lung epithelial cells (A549) were compared by in vitro toxicity assays and chemical component analysis. Both air levels and cytotoxic effects of PM2.5 varied with the transition of haze event. Although the concentration of PM2.5 in air is of course the highest in pollution stage driven by local stable meteorological condition, unit mass of them posed higher toxicity (lower cell viability and higher IL-6) but induced lower cell oxidative (evidences of ROS and NQO1 mRNA expression) and inflammatory cytokine TNF-α responses than those particles during non-pollution stage. These patterns were explained by the metals and water-soluble components decreased with the haze development. Non-soluble particulate carbonaceous aerosol compositions might play a significant role in inducing cytotoxicity. Moreover, the regional pattern of episode pollution weakened the spatial variation within a city scale. Since the haze development intensified both the quantity and toxicity of PM2.5 in air, the health risks of overall aerosol exposure were synthetically amplified during haze weather, so the increased air particles with higher toxic components from fuel combustion sources should be key targets of pollution control.